This dataset contains the code and data necessary to reproduce the results from the paper "Experimental Evaluation of Individualized Treatment Rules".

SYNTHETIC EXPERIMENTS

For the synthetic experiments, Xt.csv is the master data used as the population for the synthetic experiments. 

generate_testing_data.R is the master helper file to generate the synthetic data based on high/low treatment effect.

BARTTest.R calculates the various statistics shown in the paper for Bayesian Additive Regression Trees using the synthetic data. 

CFTest.R calculates the various statistics shown in the paper for Causal Forests using the synthetic data. 

LASSOcvTest calculates the various statistics shown in the paper for LASSO using the synthetic data (and in particular, also includes the experiments under the cross-validation setting). 

REAL-WORLD EXPERIMENTS

For the real-world experiments, the master datasets used are the STAR_students.sav file and STAR_K-3_Schools.sav file. 

RealExp.R is the file used to process the data in the fixed case to a training and testing set, which the values are then calculated in RuralSchools.R. 
To ensure repeatability, the actual training/testing set is given in limited_training_set_new.csv and limited_testing_set_new.csv.

RuralSchoolscv.R does the analysis for the cross-validation case using the master files directly. 